Online education industry course recommendation method and device based on large model
Through the online education course recommendation method based on big model, combined with user identity verification, course requirements preprocessing and deep learning models, the problem of inaccurate course recommendation in the existing technology is solved, and more efficient user requirements matching and recommendation results are achieved. Compliance improves user satisfaction and timeliness of recommendations.
Patent Information
- Application Number
- CN202411196568.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-12-06
AI Technical Summary
The existing online education course recommendation methods are difficult to accurately match the user's career background and needs, resulting in a far different recommendation results from the actual needs of users.
Using a large model-based online education industry course recommendation method, we can ensure the recommended through user authentication, course requirements preprocessing, searching course knowledge base and network resources, using deep learning models, and filtering and sorting them in combination with user history information to ensure the recommended Compliance and legality.
Improves the accuracy and user satisfaction of course recommendations, ensures that recommendation results match user needs, and maintains the timeliness of the model through regular updates to the knowledge base and user feedback.
Smart Images

Figure CN119089045A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of big model technology, and in particular to a big model-based online education industry course recommendation method and device. Background Art
[0002] With the booming development of online education, how to provide users with course recommendations that match their professional background and needs has become a key issue. When searching for corresponding course information through keywords, the course information retrieved may be far from the actual requirements of users. Matching needs only through course information no longer meets actual needs. Summary of the invention
[0003] In view of the needs and shortcomings of current technological development, the present invention provides a method and device for recommending courses in the online education industry based on a large model to improve the quality and efficiency of online course recommendations.
[0004] In the first aspect, the present invention provides a course recommendation method for the online education industry based on a large model. The technical solution adopted to solve the above technical problems is as follows:
[0005] A course recommendation method for the online education industry based on a large model, comprising the following steps:
[0006] S1. The user submits identity information and course requirement information;
[0007] S2, verifying the user's rights based on the user's identity information. If the verification fails, automatically replying to the user with a prompt message according to a preset template. If the verification succeeds, proceeding to step S3;
[0008] S3, preprocessing the user's course demand information, searching in a pre-built course knowledge base based on the preprocessing result, obtaining local course information including the course name and relevant knowledge points of the course content, and then searching online resources based on the preprocessing result to obtain online course information including the course name and relevant knowledge points of the course content;
[0009] S4, submitting the user's course demand information, local course information and / or online course information to the big model, which generates course recommendations after analysis;
[0010] S5. Based on the user's online education history information, filter and sort the course recommendations generated by the large model for the user to view;
[0011] S6. Users provide evaluation and feedback on the course recommendations generated by the big model to optimize the big model.
[0012] Optionally, a sensitive word list is pre-defined, which contains all words that violate laws, regulations or social moral standards;
[0013] After the user submits his / her identity information and course demand information, the user's course demand information is screened through a sensitive word list to ensure the compliance of the user's course demand information;
[0014] If the screening fails, the preset information will be returned and the user will be prompted to modify the course requirement information;
[0015] If the user passes the screening, the third-party security interface will be called to verify the legitimacy of the user's course demand information.
[0016] If the verification fails, the preset rejection message will be returned, and the user will be prompted to comply with laws, regulations and social moral standards;
[0017] If the verification passes, continue to step S3.
[0018] Further optionally, for the recommended courses generated by the large model, the content of the recommended courses is first screened through a list of sensitive words to ensure the compliance of the content. If it fails the screening, the preset information is returned and the recommended course is deleted. If it passes the screening, the third-party security interface is then called to perform a legality check on the content of the recommended course. If the check fails, the process ends and a preset rejection message is returned, with a prompt that the recommended course does not comply with laws, regulations and social ethical standards. If the check passes, the recommended course is returned for the user to view.
[0019] Optionally, step S3 is performed to preprocess the user course demand information, and based on the preprocessing result, a search is performed in a pre-built course knowledge base to obtain local course information including the course name and knowledge points related to the course content, specifically including:
[0020] S3.1. Use natural language processing technology to pre-process user course demand information by standardizing, segmenting, and correcting text errors;
[0021] S3.2. Generate a search query based on the word segmentation results after text error correction, search the course knowledge base through the Elasticsearch search engine, and obtain preliminary search results;
[0022] S3.3. If the preliminary search result meets the set threshold, the preliminary search result is output as the local final search result;
[0023] S3.4, if the preliminary search result does not meet the set threshold, further execute step S3.5;
[0024] S3.5, use the bge search method to search the course knowledge base and obtain the secondary search results;
[0025] S3.6. If the secondary search result meets the set threshold, the secondary search result is output as the local final search result;
[0026] S3.7. If the secondary search result does not meet the set threshold, the preliminary search result and the secondary search result of step S3.4 are superimposed as the local final search result.
[0027] Optionally, the large model is a deep learning model trained based on historical education data to understand user needs and provide course recommendations. The historical education data includes course names, relevant knowledge points in the course content, the industry field to which the course belongs, and user reviews;
[0028] The training and optimization process of the large model is as follows:
[0029] Regularly collect user evaluation feedback information and generate input-output pairs for training large models;
[0030] Collect and update course information in the course knowledge base and online resources, and perform data cleaning, deduplication and preprocessing operations on the collected course information;
[0031] The course information after preprocessing is randomly divided into training set, validation set and test set;
[0032] According to the specific data and tuning goals, set the training strategy that affects the training speed and performance of the large model, and use the training set to train the large model; after setting the training set to iteratively update the model parameters, use the validation set to fine-tune the large model; use the test set to evaluate the performance of the fine-tuned large model to meet the expected tuning goals;
[0033] Deploy the large model that meets expectations into the application environment and make course recommendations.
[0034] In a second aspect, the present invention provides a course recommendation device for the online education industry based on a large model, and the technical solution adopted to solve the above technical problems is as follows:
[0035] A device for recommending courses in the online education industry based on a large model, comprising the following modules:
[0036] Information submission module, used to assist users in submitting identity information and course requirement information;
[0037] The identity authentication module is used to verify the user's authority based on the user's identity information. If the verification fails, the user will be automatically replied with a prompt message according to the preset template. If the verification succeeds, the user's course demand information will be sent to the information preprocessing module;
[0038] Information preprocessing module, used to preprocess user course demand information;
[0039] The course search module is used to search in a pre-built course knowledge base based on the pre-processing results to obtain local course information including course names and knowledge points related to course content, and to search online resources based on the pre-processing results to obtain online course information including course names and knowledge points related to course content;
[0040] A result submission module, used to submit user course demand information, local course information and / or online course information to the big model;
[0041] The big model is used to analyze the submitted user course demand information, local course information and / or online course information to generate course recommendations;
[0042] The result filtering module is used to filter and sort the course recommendations generated by the large model based on the user's online education history information for users to view;
[0043] The display feedback module is used to assist users in viewing the course recommendations output by the result filtering module and to provide evaluation feedback on the course recommendation results to achieve optimization of the large model.
[0044] Optionally, the online education industry course recommendation device may further include:
[0045] A predefined module is used to predefine a sensitive word list, which contains all words that violate laws, regulations or social moral standards;
[0046] The compliance screening module is used to screen the user's course demand information through a sensitive word list after the user submits his / her identity information and course demand information to ensure the compliance of the user's course demand information;
[0047] The screening processing module is used to return the preset information when the information submitted by the user fails to pass the screening, and prompt the user to modify the course demand information. When the information submitted by the user passes the screening, the third-party security interface is called to verify the legitimacy of the user's course demand information;
[0048] The verification processing module is used to return the preset rejection information when the user's course demand information fails, and prompt the user to comply with laws, regulations and social moral standards. The information preprocessing module is called when the user's course demand information passes the verification.
[0049] Optionally, after the large model involved generates recommended courses, the compliance screening module first screens the content of the recommended courses through a list of sensitive words to ensure the compliance of the content. Secondly, the screening processing module returns preset information when the content of the recommended course fails the screening and deletes the recommended course. When the content of the recommended course passes the screening, the third-party security interface is called to verify the legality of the content of the recommended course. Finally, the verification processing module ends the process when the content verification of the recommended course fails, returns the preset rejection information, and prompts that the recommended course does not comply with laws, regulations and social ethical standards. When the content verification of the recommended course passes, the recommended course is returned for the user to view later.
[0050] Optionally, the information preprocessing module involved uses natural language processing technology to perform preprocessing operations such as standardization, word segmentation and text error correction on the user course demand information;
[0051] The course search module searches the pre-built course knowledge base based on the pre-processing results to obtain local course information including course names and relevant knowledge points of course content. The specific process is as follows:
[0052] Through natural language processing technology, the user's course demand information is pre-processed by standardization, word segmentation and text error correction;
[0053] Generate a search query based on the word segmentation results after text error correction, search the course knowledge base through the Elasticsearch search engine, and obtain preliminary search results;
[0054] If the preliminary search result meets the set threshold, the preliminary search result is output as the local final search result;
[0055] If the initial search results do not meet the set threshold, the bge search method is further used to search the course knowledge base to obtain secondary search results;
[0056] If the secondary search result meets the set threshold, the secondary search result is output as the local final search result;
[0057] If the secondary search result does not meet the set threshold, the preliminary search result and the secondary search result of step S3.4 are superimposed as the local final search result.
[0058] Optionally, the large model involved is a deep learning model trained based on historical education data for understanding user needs and providing course recommendations. The historical education data includes course names, relevant knowledge points in the course content, the industry field to which the course belongs, and user evaluations;
[0059] The training and optimization process of the large model is as follows:
[0060] Regularly collect user evaluation feedback information and generate input-output pairs for training large models;
[0061] Collect and update course information in the course knowledge base and online resources, and perform data cleaning, deduplication and preprocessing operations on the collected course information;
[0062] The course information after preprocessing is randomly divided into training set, validation set and test set;
[0063] According to the specific data and tuning goals, set the training strategy that affects the training speed and performance of the large model, and use the training set to train the large model; after setting the training set to iteratively update the model parameters, use the validation set to fine-tune the large model; use the test set to evaluate the performance of the fine-tuned large model to meet the expected tuning goals;
[0064] Deploy the large model that meets expectations into the application environment and make course recommendations.
[0065] The method and device for recommending courses in the online education industry based on a large model of the present invention have the following beneficial effects compared with the prior art:
[0066] 1. The present invention can effectively integrate the course knowledge base and the knowledge question-answering ability of the large model, and improve the accuracy of course recommendations with the help of external online retrieval; at the same time, combined with the user's online education history training information, the recommended courses are filtered again to further improve the user satisfaction with course recommendations;
[0067] 2. The present invention can regularly collect and update course information in the course knowledge base and network resources, combine user feedback information, and add new course information and data, so as to maintain the timeliness and knowledge coverage of the large model. BRIEF DESCRIPTION OF THE DRAWINGS
[0068] Attached Figure 1 is a flow chart of a method according to Embodiment 1 of the present invention;
[0069] Attached Figure 2 is a flowchart of searching the course knowledge base in step S3 in the first embodiment of the present invention;
[0070] Attached Figure 3 is a module connection block diagram of the second embodiment of the present invention;
[0071] Attached Figure 4 It is a flowchart of iterative optimization of a large model in an embodiment of the present invention. DETAILED DESCRIPTION
[0072] In order to make the technical solution, the technical problem solved and the technical effect of the present invention more clearly understood, the technical solution of the present invention is clearly and completely described below in conjunction with specific embodiments.
[0073] Embodiment 1:
[0074] Combined with Figure 1 This embodiment proposes a course recommendation method for the online education industry based on a large model, which includes the following steps:
[0075] S1. The user submits identity information and course requirement information.
[0076] S2. Verify user rights based on user identity information. If verification fails, automatically reply to the user with a prompt message according to the preset template. If verification succeeds, continue to verify the compliance and legality of the user's course demand information, including:
[0077] Predefine a sensitive word list, which contains all words that violate laws, regulations or social moral standards;
[0078] After the user submits his / her identity information and course demand information, the user's course demand information is screened through a sensitive word list to ensure the compliance of the user's course demand information;
[0079] If the screening fails, the preset information will be returned and the user will be prompted to modify the course requirement information;
[0080] If the user passes the screening, the third-party security interface will be called to verify the legitimacy of the user's course demand information.
[0081] If the verification fails, the preset rejection message will be returned, and the user will be prompted to comply with laws, regulations and social moral standards;
[0082] If the verification passes, continue to step S3.
[0083] S3. Preprocess the user's course demand information, search in a pre-built course knowledge base based on the preprocessing results to obtain local course information including the course name and knowledge points related to the course content, and then search online resources based on the preprocessing results to obtain online course information including the course name and knowledge points related to the course content.
[0084] (1) This step preprocesses the user's course demand information and searches the pre-built course knowledge base based on the preprocessing results to obtain local course information including course names and relevant knowledge points of course content. Figure 2 , specifically including:
[0085] S3.1. Use natural language processing technology to pre-process user course demand information by standardizing, segmenting, and correcting text errors;
[0086] S3.2. Generate a search query based on the word segmentation results after text error correction, search the course knowledge base through the Elasticsearch search engine, and obtain preliminary search results;
[0087] S3.3. If the preliminary search result meets the set threshold, the preliminary search result is output as the local final search result;
[0088] S3.4, if the preliminary search result does not meet the set threshold, further execute step S3.5;
[0089] S3.5, use the bge search method to search the course knowledge base and obtain the secondary search results;
[0090] S3.6. If the secondary search result meets the set threshold, the secondary search result is output as the local final search result;
[0091] S3.7. If the secondary search result does not meet the set threshold, the preliminary search result and the secondary search result of step S3.4 are superimposed as the local final search result.
[0092] (2) This step uses natural language processing technology to perform preprocessing operations such as standardization, word segmentation, and text error correction on the user's course demand information. Then, based on the preprocessing results, an online search for network resources is performed, such as searching on pre-specified online education platforms Coursera, Udemy, and edX, or using academic search engines such as Google Scholar and Baidu Scholar to obtain online course information containing course names and knowledge points related to the course content.
[0093] S4. Submit the user's course demand information, local course information and / or online course information to the big model, which generates course recommendations after analysis.
[0094] For the recommended courses generated by the large model, the content of the recommended courses is first screened through a list of sensitive words to ensure the compliance of the content. If it fails the screening, the preset information is returned and the recommended course is deleted. If it passes the screening, the third-party security interface is then called to verify the legality of the content of the recommended course. If the verification fails, the process ends and the preset rejection information is returned, with a prompt that the recommended course does not comply with laws, regulations and social ethical standards. If the verification passes, the recommended course is returned for the user to view.
[0095] S5. Based on the user's online education history information, filter and sort the course recommendations generated by the large model for the user to view.
[0096] S6. Users provide evaluation and feedback on the course recommendations generated by the big model to optimize the big model.
[0097] In addition to the above steps, the big model is a deep learning model trained based on historical education data to understand user needs and provide course recommendations. The historical education data includes course names, relevant knowledge points in the course content, the industry field to which the course belongs, and user reviews.
[0098] Combined with Figure 4 , the training and optimization process of the large model is as follows:
[0099] Regularly collect user evaluation feedback information and generate input-output pairs for training large models;
[0100] Collect and update course information in the course knowledge base and online resources, and perform data cleaning, deduplication and preprocessing operations on the collected course information;
[0101] The course information after preprocessing is randomly divided into training set, validation set and test set;
[0102] According to the specific data and tuning goals, set the training strategy that affects the training speed and performance of the large model, and use the training set to train the large model; after setting the training set to iteratively update the model parameters, use the validation set to fine-tune the large model; use the test set to evaluate the performance of the fine-tuned large model to meet the expected tuning goals;
[0103] Deploy the large model that meets expectations into the application environment and make course recommendations.
[0104] Embodiment 2:
[0105] Combined with Figure 3 This embodiment proposes a course recommendation device for the online education industry based on a large model, which includes the following modules:
[0106] Information submission module, used to assist users in submitting identity information and course requirement information;
[0107] The identity authentication module is used to verify the user's authority based on the user's identity information. If the verification fails, the user will be automatically replied with a prompt message according to the preset template. If the verification succeeds, the user's course demand information will be sent to the information preprocessing module;
[0108] Information preprocessing module, used to preprocess user course demand information;
[0109] The course search module is used to search in a pre-built course knowledge base based on the pre-processing results to obtain local course information including course names and knowledge points related to course content, and to search online resources based on the pre-processing results to obtain online course information including course names and knowledge points related to course content;
[0110] A result submission module, used to submit user course demand information, local course information and / or online course information to the big model;
[0111] The big model is used to analyze the submitted user course demand information, local course information and / or online course information to generate course recommendations;
[0112] The result filtering module is used to filter and sort the course recommendations generated by the large model based on the user's online education history information for users to view;
[0113] The display feedback module is used to assist users in viewing the course recommendations output by the result filtering module and to provide evaluation feedback on the course recommendation results to achieve optimization of the large model.
[0114] In this embodiment, the online education industry course recommendation device also includes:
[0115] A predefined module is used to predefine a sensitive word list, which contains all words that violate laws, regulations or social moral standards;
[0116] The compliance screening module is used to screen the user's course demand information through a sensitive word list after the user submits his / her identity information and course demand information to ensure the compliance of the user's course demand information;
[0117] The screening processing module is used to return the preset information when the information submitted by the user fails to pass the screening, and prompt the user to modify the course demand information. When the information submitted by the user passes the screening, the third-party security interface is called to verify the legitimacy of the user's course demand information;
[0118] The verification processing module is used to return the preset rejection information when the user's course demand information fails, and prompt the user to comply with laws, regulations and social moral standards. The information preprocessing module is called when the user's course demand information passes the verification.
[0119] It needs to be further explained that after the big model generates a recommended course, the compliance screening module first screens the content of the recommended course through a list of sensitive words to ensure the compliance of the content. Secondly, the screening processing module returns preset information when the content of the recommended course fails the screening and deletes the recommended course. When the content of the recommended course passes the screening, the third-party security interface is called to verify the legality of the content of the recommended course. Finally, the verification processing module ends the process when the content verification of the recommended course fails, returns the preset rejection information, and prompts that the recommended course does not comply with laws, regulations and social ethical standards. When the content verification of the recommended course passes, the recommended course is returned for the user to view later.
[0120] In this embodiment, the information preprocessing module uses natural language processing technology to perform preprocessing operations such as standardization, word segmentation and text error correction on the user course demand information.
[0121] The course search module searches the pre-built course knowledge base based on the pre-processing results to obtain local course information including course names and relevant knowledge points of course content. Figure 2 , the specific process is as follows:
[0122] Through natural language processing technology, the user's course demand information is pre-processed by standardization, word segmentation and text error correction;
[0123] Generate a search query based on the word segmentation results after text error correction, search the course knowledge base through the Elasticsearch search engine, and obtain preliminary search results;
[0124] If the preliminary search result meets the set threshold, the preliminary search result is output as the local final search result;
[0125] If the initial search results do not meet the set threshold, the bge search method is further used to search the course knowledge base to obtain secondary search results;
[0126] If the secondary search result meets the set threshold, the secondary search result is output as the local final search result;
[0127] If the secondary search result does not meet the set threshold, the preliminary search result and the secondary search result of step S3.4 are superimposed as the local final search result.
[0128] The course search module searches for online resources based on the preprocessing results to obtain online course information including course names and knowledge points related to the course content. In this process, the online resources searched can be pre-specified online education platforms Coursera, Udemy, edX, or academic search engines such as Google Scholar and Baidu Scholar can be used to search for online resources.
[0129] The large model involved in this embodiment is a deep learning model trained based on historical education data for understanding user needs and providing course recommendations. The historical education data includes course names, relevant knowledge points in the course content, the industry field to which the course belongs, and user evaluations;
[0130] Combined with Figure 4 , the training and optimization process of the large model is as follows:
[0131] Regularly collect user evaluation feedback information and generate input-output pairs for training large models;
[0132] Collect and update course information in the course knowledge base and online resources, and perform data cleaning, deduplication and preprocessing operations on the collected course information;
[0133] The course information after preprocessing is randomly divided into training set, validation set and test set;
[0134] According to the specific data and tuning goals, set the training strategy that affects the training speed and performance of the large model, and use the training set to train the large model; after setting the training set to iteratively update the model parameters, use the validation set to fine-tune the large model; use the test set to evaluate the performance of the fine-tuned large model to meet the expected tuning goals;
[0135] Deploy the large model that meets expectations into the application environment and make course recommendations.
[0136] In summary, the method and device for recommending courses in the online education industry based on a big model of the present invention can effectively integrate the course knowledge base with the knowledge question-answering ability of the big model, improve the accuracy of course recommendations, and improve user satisfaction with course recommendations.
[0137] The above specific examples are used to explain the principles and implementation methods of the present invention in detail. These examples are only used to help understand the core technical content of the present invention. Based on the above specific embodiments of the present invention, any improvements and modifications made by technicians in this technical field without departing from the principles of the present invention should fall within the scope of patent protection of the present invention.
Claims
1. A course recommendation method for online education industry based on big model, It is characterized in that The steps include: S1. The user submits identity information and course requirement information; S2, verifying the user's rights based on the user's identity information. If the verification fails, automatically replying to the user with a prompt message according to a preset template. If the verification succeeds, proceeding to step S3; S3, preprocessing the user's course demand information, searching in a pre-built course knowledge base based on the preprocessing result, obtaining local course information including the course name and relevant knowledge points of the course content, and then searching online resources based on the preprocessing result to obtain online course information including the course name and relevant knowledge points of the course content; S4, submitting the user's course demand information, local course information and / or online course information to the big model, which generates course recommendations after analysis; S5. Based on the user's online education history information, filter and sort the course recommendations generated by the large model for the user to view; S6. Users provide evaluation and feedback on the course recommendations generated by the big model to optimize the big model.
2. According to claim 1, a method for recommending courses in the online education industry based on a large model, It is characterized in that Predefine a sensitive word list, which contains all words that violate laws, regulations or social moral standards; After the user submits his / her identity information and course demand information, the user's course demand information is screened through a sensitive word list to ensure the compliance of the user's course demand information; If the screening fails, the preset information will be returned and the user will be prompted to modify the course requirement information; If the user passes the screening, the third-party security interface will be called to verify the legitimacy of the user's course demand information. If the verification fails, the preset rejection message will be returned, and the user will be prompted to comply with laws, regulations and social moral standards; If the verification passes, continue to step S3.
3. According to claim 2, a method for recommending courses in the online education industry based on a large model, It is characterized in that For the recommended courses generated by the large model, the content of the recommended courses is first screened through a list of sensitive words to ensure the compliance of the content. If it fails the screening, the preset information is returned and the recommended course is deleted. If it passes the screening, the third-party security interface is then called to verify the legality of the content of the recommended course. If the verification fails, the process ends and the preset rejection information is returned, with a prompt that the recommended course does not comply with laws, regulations and social ethical standards. If the verification passes, the recommended course is returned for the user to view.
4. According to claim 1, a method for recommending courses in the online education industry based on a large model, It is characterized in that Execute step S3 to preprocess the user course demand information, and search in the pre-built course knowledge base based on the preprocessing result to obtain local course information including the course name and relevant knowledge points of the course content, specifically including: S3.
1. Use natural language processing technology to pre-process user course demand information by standardizing, segmenting, and correcting text errors; S3.
2. Generate a search query based on the word segmentation results after text error correction, search the course knowledge base through the Elasticsearch search engine, and obtain preliminary search results; S3.
3. If the preliminary search result meets the set threshold, the preliminary search result is output as the local final search result; S3.4, if the preliminary search result does not meet the set threshold, further execute step S3.5; S3.5, use the bge search method to search the course knowledge base and obtain the secondary search results; S3.
6. If the secondary search result meets the set threshold, the secondary search result is output as the local final search result; S3.
7. If the secondary search result does not meet the set threshold, the preliminary search result and the secondary search result of step S3.4 are superimposed as the local final search result.
5. According to claim 1, a method for recommending courses in the online education industry based on a large model, It is characterized in that The large model is a deep learning model trained based on historical education data for understanding user needs and providing course recommendations. The historical education data includes course names, knowledge points related to course content, industry fields to which the courses belong, and user reviews; The training and optimization process of the large model is as follows: Regularly collect user evaluation feedback information and generate input-output pairs for training large models; Collect and update course information in the course knowledge base and online resources, and perform data cleaning, deduplication and preprocessing operations on the collected course information; The course information after preprocessing is randomly divided into training set, validation set and test set; According to the specific data and tuning goals, set the training strategy that affects the training speed and performance of the large model, and use the training set to train the large model; after setting the training set to iteratively update the model parameters, use the validation set to fine-tune the large model; Use the test set to evaluate the performance of the fine-tuned large model to meet the expected tuning goals; Deploy the large model that meets expectations into the application environment and make course recommendations.
6. A course recommendation device for online education industry based on a large model, It is characterized in that It includes the following modules: Information submission module, used to assist users in submitting identity information and course requirement information; The identity authentication module is used to verify the user's authority based on the user's identity information. If the verification fails, the user will be automatically replied with a prompt message according to the preset template. If the verification succeeds, the user's course demand information will be sent to the information preprocessing module; Information preprocessing module, used to preprocess user course demand information; The course search module is used to search in a pre-built course knowledge base based on the pre-processing results to obtain local course information including course names and knowledge points related to course content, and to search online resources based on the pre-processing results to obtain online course information including course names and knowledge points related to course content; A result submission module, used to submit user course demand information, local course information and / or online course information to the big model; The big model is used to analyze the submitted user course demand information, local course information and / or online course information to generate course recommendations; The result filtering module is used to filter and sort the course recommendations generated by the large model based on the user's online education history information for users to view; The display feedback module is used to assist users in viewing the course recommendations output by the result filtering module and to provide evaluation feedback on the course recommendation results to achieve optimization of the large model.
7. The device for recommending courses in the online education industry based on a large model according to claim 6, It is characterized in that The device also includes: A predefined module is used to predefine a sensitive word list, which contains all words that violate laws, regulations or social moral standards; The compliance screening module is used to screen the user's course demand information through a sensitive word list after the user submits his / her identity information and course demand information to ensure the compliance of the user's course demand information; The screening processing module is used to return the preset information when the information submitted by the user fails to pass the screening, and prompt the user to modify the course demand information. When the information submitted by the user passes the screening, the third-party security interface is called to verify the legitimacy of the user's course demand information; The verification processing module is used to return the preset rejection information when the user's course demand information fails, and prompt the user to comply with laws, regulations and social moral standards. The information preprocessing module is called when the user's course demand information passes the verification.
8. The device for recommending courses in the online education industry based on a large model according to claim 7, It is characterized in that After the big model generates recommended courses, the compliance screening module first screens the content of the recommended courses through a list of sensitive words to ensure the compliance of the content. Secondly, the screening processing module returns preset information when the content of the recommended course fails the screening and deletes the recommended course. When the content of the recommended course passes the screening, the third-party security interface is called to verify the legality of the content of the recommended course. Finally, the verification processing module ends the process when the content verification of the recommended course fails, returns the preset rejection information, and prompts that the recommended course does not comply with laws, regulations and social ethical standards. When the content verification of the recommended course passes, the recommended course is returned for users to view later.
9. The device for recommending courses in the online education industry based on a large model according to claim 6, It is characterized in that The information preprocessing module uses natural language processing technology to perform preprocessing operations such as standardization, word segmentation and text error correction on user course demand information; The course search module searches in the pre-built course knowledge base based on the pre-processing results to obtain local course information including course names and knowledge points related to course content. The specific process is as follows: Through natural language processing technology, the user's course demand information is pre-processed by standardization, word segmentation and text error correction; Generate a search query based on the word segmentation results after text error correction, search the course knowledge base through the Elasticsearch search engine, and obtain preliminary search results; If the preliminary search result meets the set threshold, the preliminary search result is output as the local final search result; If the initial search results do not meet the set threshold, the bge search method is further used to search the course knowledge base to obtain secondary search results; If the secondary search result meets the set threshold, the secondary search result is output as the local final search result; If the secondary search result does not meet the set threshold, the preliminary search result and the secondary search result of step S3.4 are superimposed as the local final search result.
10. The device for recommending courses in the online education industry based on a large model according to claim 6, It is characterized in that The large model is a deep learning model trained based on historical education data for understanding user needs and providing course recommendations. The historical education data includes course names, knowledge points related to course content, industry fields to which the courses belong, and user reviews; The training and optimization process of the large model is as follows: Regularly collect user evaluation feedback information and generate input-output pairs for training large models; Collect and update course information in the course knowledge base and online resources, and perform data cleaning, deduplication and preprocessing operations on the collected course information; The course information after preprocessing is randomly divided into training set, validation set and test set; According to the specific data and tuning goals, set the training strategy that affects the training speed and performance of the large model, and use the training set to train the large model; after setting the training set to iteratively update the model parameters, use the validation set to fine-tune the large model; Use the test set to evaluate the performance of the fine-tuned large model to meet the expected tuning goals; Deploy the large model that meets expectations into the application environment and make course recommendations.
Citation Information
Patent Citations
Online course recommendation method and system
CN115146162A
Online education course request processing method based on artificial intelligence
CN118132850A
Cited By
Intelligent online video teaching effect management system
CN119672764A
Intelligent online video teaching effect management system
CN119672764B
User course recommendation method and system based on large model generation
CN121388283A
